Date: Tue, 13 Mar 2012 10:02:16 +0000 (UTC) From: jb <jb.1234abcd@gmail.com> To: freebsd-questions@freebsd.org Subject: Re: lost+found dir placement Message-ID: <loom.20120313T105734-220@post.gmane.org> References: <loom.20120313T085550-787@post.gmane.org> <201203130825.q2D8Pa6Y053252@mail.r-bonomi.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Robert Bonomi <bonomi <at> mail.r-bonomi.com> writes: > ... > The fsck_ffs manpage says that 'lost+found' is _created_ *when*needed*, > in the root of a filesystem, if not already present. > > The presense of /mnt/lost+found is _not_ an error. just a surperfluous > file that ended up there 'somehow'. > ... This worried me. And still does ... > *IF* you're going to file a PR, it should be for the filesystem > initialization process -- which "should" (a) create the lost+found > directory, (b) create some 'reasonable' number of files in that directory, > and (c) then delete all those files. This ensures that the directory > exists and has disk-space allocated for a 'reasonable' number of > 'recovered' file entries. > That's perhaps why under Linux they have special mklost+found entry ? > The existing fsck_ffs has a catastrophic failure mode if there is no > space on the disk for the lost+found directory to grow to acomodate > the recovered file entries. > I was surprised to find empty lost+found dir in /mnt. drwx------ 2 root wheel 512 May 5 2011 lost+found That's why I jumped a bit. Few days ago, after clean reboot to single user mode, I tested fsck manually on SUJ fs and found things that seemed to be questionable (I posted it on current@ list, if you want to take a look). So, it must have happened during that time, because as I said I did not have any forced fsck run at boot times, and I almost swear I did not have this lost+found dir in /mnt before. I will take a look at source code of fsck* entries and perhaps find a clue. jb
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?loom.20120313T105734-220>